cd /news/ai-products/benchmark-for-ai-assistants ยท home โ€บ topics โ€บ ai-products โ€บ article
[ARTICLE ยท art-129931] src=assistantbenchmark.com โ†— pub= topic=ai-products verified=true sentiment=ยท neutral

Benchmark for AI Assistants

A new scorecard benchmarks 71 AI assistants across 15 dimensions on a single scale, with Muse leading the in-progress field at 9.1 after scoring 7 of 15 dimensions. Instinct ranks second at 8.4 with 11 of 15 dimensions scored, followed by szn at 8.0, while Pally and Ollie are tied at 7.6 with 13 of 15 dimensions each. No assistant has completed all 15 dimensions, and 23 remain pending with no score.

read8 min views1 publishedSep 15, 2026
Benchmark for AI Assistants
Image: source

Purchasing a productยท Partial ยท test ยท Staged a Stumptown pickup at 18 W 29th and stopped before purchase with nothing charged.

71 assistants, 15 dimensions, one scale. Overall is a running mean until an assistant is fully tested.

Assistant Tested
Completed0None yet. An assistant lands here once every applicable dimension is scored.
In progress21Some dimensions scored. Overall is a partial mean and will move as tests land.
Muse 7s 9.1 7/15 9 7 9 10 10 โ€” โ€” 10 9 โ€” โ€” โ€” โ€” โ€” โ€”
Instinct 18s 8.4 11/15 8 10 9 9 9 8 9 8 5 9 โ€” โ€” โ€” 8 โ€”
szn 39s 8.0 11/15 8 8 7 8 8 โ€” โ€” 7 7 10 10 โ€” 7 โ€” 8
Pally โ€” 7.6 13/15 7 8 9 8 8 8 โ€” 7 8 โ€” 4 7 9 7 9
Ollie โ€” 7.6 13/15 9 7 7 8 8 8 โ€” 7 8 โ€” 4 9 8 7 9
Tomo 11s 7.6 10/14 7 9 8 8 8 โ€” โ€” 7 7 6 โ€” N/A 7 โ€” 9
Asaply โ€” 7.5 2/11 โ€” N/A 7 8 N/A โ€” โ€” โ€” โ€” โ€” N/A โ€” โ€” โ€” N/A
Shuffle 55s 7.4 11/14 7.5 9 9 8 8 โ€” โ€” 7 7 6 4 N/A 7 โ€” 9
Grok Bot 10s 7.3 7/15 8 โ€” 7 8 9 โ€” โ€” 3 โ€” โ€” โ€” โ€” 8 โ€” 8
Caddy 32s 7.2 10/14 8 8 7 7 7 โ€” โ€” 7 โ€” 10 3 N/A 7 โ€” 8
Catch 48s 6.5 2/15 6 โ€” โ€” โ€” โ€” โ€” 7 โ€” โ€” โ€” โ€” โ€” โ€” โ€” โ€”
Poke 12s 6.3 10/15 3 โ€” 8 โ€” 8 7 7 7 โ€” 8 โ€” 5 8 2 โ€”
Wajo / Fo 33s 6.2 6/15 2 7 7 โ€” 6 โ€” 7 โ€” โ€” 8 โ€” โ€” โ€” โ€” โ€”
Asmi โ€” 6.2 13/15 6 5 7 5 8 5 โ€” 6 8 โ€” 4 3 8 7 9
Boba โ€” 6.1 8/15 6 5 8 5 8 โ€” โ€” โ€” โ€” 6 4 โ€” โ€” โ€” 7
Folk โ€” 6.0 3/15 6 6 6 โ€” โ€” โ€” โ€” โ€” โ€” โ€” โ€” โ€” โ€” โ€” โ€”
Boski 9s 5.3 4/15 3 โ€” 4 โ€” โ€” โ€” โ€” โ€” 6 โ€” โ€” โ€” โ€” โ€” 8
tinyNature 9s 5.0 6/15 4 โ€” 5 โ€” 6 7 5 โ€” โ€” 3 โ€” โ€” โ€” โ€” โ€”
Brea 7s 5.0 4/15 3 โ€” 7 โ€” โ€” โ€” โ€” โ€” 6 โ€” โ€” โ€” โ€” โ€” 4
Town 12s 3.3 3/15 โ€” โ€” โ€” โ€” โ€” 6 2 2 โ€” โ€” โ€” โ€” โ€” โ€” โ€”
OpenInstinct 23s 1.0 1/15 โ€” โ€” โ€” โ€” โ€” โ€” 1 โ€” โ€” โ€” โ€” โ€” โ€” โ€” โ€”
Pending23On the roster, not tested yet. No score.

9 6 2 score11s median replyโ€” not testedN/A doesn't applyHow scoring works

Latest #

New tests and quotes. Responding to emailsยท N/A ยท test ยท Declined email replies as outside its coffee, food, and local-spots scope.

Recommendation qualityยท Partial ยท test ยท Recommended Paper Coffee, The Coffee Shop, and Small Space Coffee near 800 6th under $20 and independent after chain noise.

Phone callsยท N/A ยท test ยท Declined outbound calling as outside its coffee, food, and local-spots scope.

@grokbot is getting more useful by the day... For example, I just had it scan my email for anything that looks like a bill. Then I gave it a CSV from my bank account and creditโ€ฆ

Grok Bot just spun up the army of four: Research, Options, Book, Killer. One finds ideas. One builds tickets. One watches the money. One tries to kill bad trades before I click.

Instinct giving contextual product updates is one of the coolest things I've ever seen > knew I was looking for US Open tickets last week > used that info to tell me next time it will be able to coordinate with my friends' agent to handle it Amazing touch @noahrshinn

The 11 Instinct AI @noahrshinn prompts I've found most helpful - steal than, and comment with any you recommend. ๐Ÿญ. ๐— ๐—ฒ๐—ฑ๐—ถ๐—ฐ๐—ฎ๐—น ๐—ฏ๐—ถ๐—น๐—น๐˜€. "I have [insurer]. Pull my EOBs and annual statement, reconcile every claim, flag anything denied or billed above patient responsibility, and draft the appeals. Send nothing without my approval." ๐Ÿฎ. ๐—”๐—บ๐—ฎ๐˜‡๐—ผ๐—ป ๐—ฎ๐˜‚๐—ฑ๐—ถ๐˜. "Audit my full order history for refunds I never received, price drops, and double charges. Recover what you can." ๐Ÿฏ. ๐—™๐—น๐—ถ๐—ด๐—ต๐˜ ๐—ฑ๐—ฒ๐—น๐—ฎ๐˜† ๐—บ๐—ผ๐—ป๐—ฒ๐˜†. "Flight [number] on [date] was delayed [X] hours. Check whether I'm owed compensation through the airline, travel insurance, or my credit card, and file for everyone on the booking." (This one alone was $1,200.) ๐Ÿฐ. ๐—ฆ๐˜‚๐—ฏ๐˜€๐—ฐ๐—ฟ๐—ถ๐—ฝ๐˜๐—ถ๐—ผ๐—ป ๐—ฐ๐—น๐—ฒ๐—ฎ๐—ป๐˜‚๐—ฝ. "Audit my card statements for every recurring charge, check last use date, and give me a cancel/downgrade list before touching anything." ๐Ÿฑ. ๐— ๐—ถ๐˜€๐˜€๐—ถ๐—ป๐—ด ๐—บ๐—ถ๐—น๐—ฒ๐˜€. "Check my [airline] account for miles from past flights that never posted and file the claims." ๐Ÿฒ. ๐—ฆ๐—ฝ๐—ฎ๐—บ ๐—ฐ๐—ฎ๐—น๐—น๐˜€. "Register me on the national and state Do Not Call lists, set up every spam blocker my carrier and iPhone offer, then find my personal info on people-search and data broker sites and get it removed." ๐Ÿณ. ๐—ฅ๐—ฒ๐˜€๐˜๐—ฎ๐˜‚๐—ฟ๐—ฎ๐—ป๐˜ ๐˜„๐—ฎ๐˜๐—ฐ๐—ต. "Watch [restaurant] for a table for [party size] Friday or Saturday and book anything between 8 and 9pm." ๐Ÿด. ๐—ฃ๐—ฟ๐—ถ๐—ฐ๐—ฒ ๐—ต๐˜‚๐—ป๐˜๐—ถ๐—ป๐—ด. "Find [exact product] at the cheapest delivered price to my home. Take it to the checkout screen so I can see the real total, then text me to confirm." ๐Ÿต. ๐—•๐—ถ๐—น๐—น ๐—ฟ๐—ฒ๐˜๐—ฒ๐—ป๐˜๐—ถ๐—ผ๐—ป. "Audit my [internet/wireless] bill, find the fees and rate increases, then run the retention play โ€” threaten to cancel, push until you've got their best offer, and bring it to me before accepting. Do not cancel." ๐Ÿญ๐Ÿฌ. ๐—Ÿ๐˜‚๐—ป๐—ฐ๐—ต. "Book me 1pm at the highest-rated place within walking distance with availability, [$$-$$$]." ๐Ÿญ๐Ÿญ. ๐—จ๐—ป๐—ฐ๐—น๐—ฎ๐—ถ๐—บ๐—ฒ๐—ฑ ๐—ฝ๐—ฟ๐—ผ๐—ฝ๐—ฒ๐—ฟ๐˜๐˜†. "Search every state's unclaimed property database for me and my family โ€” old accounts, refunds, deposits. My prior addresses are [X, Y]. File the claims."

โ”€โ”€ more in #ai-products 4 stories ยท sorted by recency
โ”€โ”€ more on @muse 3 stories trending now
sponsored brought to you by zahid.host 4,200+ EU-deployed projects
reading about agents? ship yours in a single git push.

Run your AI side-project on zahid.host

EU-based hosting, git-push deploys, automatic HTTPS, no cold starts. Free tier with a custom domain โ€” perfect for shipping the agent you just read about.

$git push zahid main
โ†’ Live at https://your-agent.zahid.host โœ“
Get free account โ†’ Pricing
from โ‚ฌ0/mo ยท no card required
LIVE [news/benchmark-for-ai-assโ€ฆ] indexed:0 read:8min 2026-09-15 ยท โ€”